Abstract
This research investigated the use of locally produced, non-proprietary ultra-high-performance concrete (UHPC) as a grouting material to repair deteriorated shear keys. Shear keys are used in adjacent girder superstructures to produce monolithic behavior and load transfer across the structure. Shear key degradation can jeopardize the integrity of the structure. Transportation agencies have reported that 75% of distress in adjacent girder bridges is because of cracking and de-bonding along shear keys. Previous research has shown that locally produced UHPC has excellent mechanical and durability properties. UHPC has also been shown to have good bonding characteristics that are desirable in a potential grouting material. Bond strength between UHPC grout and substrate concrete was evaluated using slant-shear and direct tension tests. Results showed that adequate bond was achieved at 7 days. Low strengths at 28 days were observed because of low strength of the substrate concrete. Shrinkage of UHPC grout was also investigated. Shrinkage at 28 days was less than 600 µstrain which is acceptable for repair practices. Full-scale testing was used to evaluate load-deflection behavior of channel girder assemblages with grouted shear keys. Results showed that UHPC grout and non-shrink grout had similar mechanical performance. Excellent bond was achieved with all grouts, even with minimal surface preparation. The similar performances of the non-shrink grout and the UHPC grout indicates that UHPC grout does not provide a mechanical benefit over the non-shrink grout.
Precast girders are commonly used in North America because of their ease of construction, affordability, favorable aesthetic appeal, and high flexural stiffness. Some girder shapes require shear keys to ensure monolithic behavior and load transfer across the superstructure. However, shear key durability is a concern to engineers since shear key degradation can jeopardize the integrity of the structure. Transportation agencies in the United States (U.S.) have reported that 75% of distress in adjacent girder bridges was because of cracking along shear keys and de-bonding of the interfaces between the girders and the shear keys ( 1 ).
This study focused on investigating the potential for using locally produced ultra-high-performance concrete (UHPC) as a grouting material to rehabilitate shear keys in adjacent pre-stressed girder bridges in New Mexico, U.S. Previous research at New Mexico State University (NMSU) has shown that locally produced UHPC has excellent mechanical and durability properties ( 2 ). UHPC has also been shown to have good bonding characteristics that are desirable in a potential grouting material ( 3 , 4 ). Additionally, the use of non-proprietary UHPC mixtures provides an economical and sustainable alternative to proprietary products. In comparison with other proprietary mixtures, locally produced UHPC has a decrease in cost of anywhere from 10% to 70% ( 5 ). To evaluate UHPC for potential use as a grout in shear keys, bond between UHPC and normal strength concrete was studied, as well as full-scale testing of UHPC grouted shear keys.
Background
UHPC became commercially available in the U.S. in 2000 ( 6 ). That version of UHPC consisted of Portland cement, ground quartz, fine sand, an accelerating admixture, a high-range water-reducing admixture (HRWRA), and 2% by volume of steel fibers ( 1 ). ASTM C1856 defines UHPC as having a compressive strength of 17,000 pounds per square inch (psi) (120 MPa) ( 7 ). Transportation agencies have reported compressive strengths greater than 150 MPa and sustained post-cracking tensile strengths greater than 720 psi (5 MPa) ( 6 ).
Shear keys are critical structural components of a bridge superstructure located between adjacent girders designed to restrain lateral displacement and facilitate load transfer between the girders. Durability of shear keys is a concern to researchers and engineers because shear key degradation can include debonding, cracking, and leaking that can compromise strength and serviceability of the bridge ( 1 ). After substantial degradation, beams will not deflect equally under live loads ( 8 ). When leakage occurs, water and salts can more easily penetrate to the reinforcement and pre-stressing strands to cause corrosion of the steel ( 9 , 10 ).
Studies have shown that UHPC has high bond strengths and that bond between the shear key and the superstructure element is affected primarily by the strength and surface preparation of the existing beam and post-tensioning across the width of the bridge ( 4 ). The minimum recommended bond strengths for repair materials for the direct tension test are 150 psi (1.0 MPa) at 7 days and 250 psi (1.7 MPa) at 28 days. For the slant-shear test, the minimum recommended shear strengths are 1000 psi (6.90 MPa) at 7 days and 2000 psi (13.8 MPa) at 28 days ( 11 ).
Methods
UHPC and Substrate Mixture
The UHPC mixture consisted of Type I/II Portland cement, silica fume, fly ash, sand passed through a No. 4 sieve, HRWRA, water, and 2% steel fibers by volume. The sand, cement, and fly ash were obtained from local sources, and the silica fume, steel fibers, and HRWRA were obtained from regional suppliers. The substrate concrete used in the laboratory investigation consisted of Type I/II Portland cement, fine and coarse aggregates, water, and air entraining admixture. Table 1 presents the mixture proportions for the UHPC and substrate concrete mixtures.
Non-Proprietary Ultra-High-Performance Concrete (UHPC) and Substrate Concrete Mixture Proportions
Note: na = not applicable.
Bond Strength Testing
Slant-shear and direct tension tests were performed to assess the bond strength between the UHPC grout and the substrate concrete.
Slant Shear
The slant-shear tests conducted for this work followed ASTM C882 ( 12 ). However, the normal strength concrete substrate was cast in a 6 by 12 in. (152 by 304 mm) cylindrical mold as opposed to a 3 by 6 in. (76 by 152 mm) cylinder mold. This modification was implemented to use commonly available molds. The increase in specimen size should result in conservatively low strengths. The specimens were moist-cured for 7 days before texturing and placement of UHPC. After de-molding, the composite specimens were cured at ambient conditions (68°F [20°C] and 30% relative humidity). The specimens were tested in compression to determine the shear strength of the bond at either 7 or 28 days.
Texture depth was determined in accordance with ASTM E965 ( 13 ). Surface preparation varied from formed to medium roughness. The formed surface was prepared by using a stiff wire brush to dust off loose material and the chipped surfaces were prepared by using an air hammer to chip off the formed surface.
The slant-shear specimens were tested in compression, as shown in Figure 1a, until failure. Three possible failure modes can occur that include failure in the UHPC overlay, failure of the bond, and failure in the substrate. Shear strength of the bond was calculated using Equation 1:
where τ = shear stress, P = ultimate load, A = cross-sectional area, and α = angle of the bonded interface from horizontal (60°).
Test setups for bond strength testing: (a) slant-shear test and (b) direct tension test.
Direct Tension
Direct tensile strength of bonded specimens was evaluated using a modified version of ASTM C1583 ( 14 ). The standard test uses a pull-off device to determine the strength from cored slabs that was replaced with the tension test apparatus shown in Figure 1b to test cylindrical cores taken from prismatic specimens.
The substrate material was cast using a 3 × 4 × 16 in. (76 × 102 × 406 mm) beam mold filled halfway with normal strength concrete. After 24 h, the substrate material was placed in a moist room (73°F [23°C] and 98% relative humidity) for 7 days to cure. After 7 days, the UHPC was applied and left to cure at ambient conditions for 7 days. Six days after the UHPC was applied, 1.875 in. (47.6 mm) diameter specimens were cored. Specimens were then bonded to end plates using epoxy. Direct tension tests were conducted 7 days and 28 days after the UHPC was applied to the substrate.
The concrete substrate was textured in the same manner as the slant-shear specimens to achieve similar texture depths.
Figure 1b illustrates the attachment of eye-bolts to the end plates epoxied to the specimen. The eye-bolts served as pivots at both ends of the specimen during loading so that concentric loading was ensured. The bond strength was calculated by using the ultimate load during testing in Equation 2:
where T = tensile strength, P = ultimate load, and Ac = cross-sectional area of the specimen.
Shrinkage and Temperature Effects
Shrinkage of the UHPC was assessed because grout shrinkage contributes to shear stresses and direct tension on the bonded interface. Shrinkage was measured according to ASTM C157 ( 15 ). The shrinkage specimens were 3 × 4 × 16 in. (76 × 102 × 406 mm) prisms with contact points cast into the ends of the specimens. After curing in the mold for 24 h, the shrinkage specimens were cured in a moist room (73°F [23°C] and 98% relative humidity) for 6 days. On day 7, the specimens were placed in ambient conditions for the remainder of shrinkage monitoring. For comparison with the wet-cured specimens, specimens were also produced and cured at ambient lab conditions for the entire duration of shrinkage monitoring. This curing modification to ASTM C157 was implemented for convenience of testing and the 30% relative humidity (ambient lab conditions) provided a harsher, more conservative environment for the UHPC specimens.
Shrinkage monitoring began immediately after the prisms were removed from the molds and continued for 28 days. Changes in length were measured using a comparator ( 15 ).
Flexural Testing
Longitudinal four-point flexural testing of two reinforced concrete channel girders connected by a grouted shear key was performed as shown in Figure 2a in the Structural Testing Laboratory at NMSU. The channel girders were recovered from a previously in-service bridge. The recovered girders were 3 ft. (0.914 mm) wide reinforced concrete channel girders with a span length of 25 ft. (7.62 m). The girder cross-section is shown in Figure 2b.
Full-scale flexural testing: (a) test setup and instrumentation and (b) channel girder cross-section.
Two girders were used to conduct three tests (one test for each grout). The first and third tests used locally developed UHPC to grout the shear key. However, the first shear key was cast with no surface preparation of the shear key to facilitate bond between the girders and the UHPC (referred to as improperly bonded). US Spec NA non-shrink grout, an approved non-shrink grout for use in New Mexico Department of Transportation (NMDOT) projects, was used to grout the shear key in the second test. Surface preparation for the test that utilized the non-shrink grout included cleaning and wetting the shear key surface, but no roughening of the surface was performed. Before casting a subsequent grouting material, the previous shear key grout was removed so that only paste residue (less than 0.02 in. [0.5 mm] in thickness) remained on the shear key surface. Cyclic loading, 1,000 cycles at service load conditions, was conducted during the two-girder shear key tests. Additionally, an individual girder (one of the previously loaded girders) was tested to quantify the load-deflection behavior of a single girder. The final test investigated the overloaded (i.e., post-yield) behavior of the system when the shear key was grouted with UHPC as shown in Table 2.
Test Descriptions
Test Setup and Instrumentation
Two reinforced concrete girders were placed adjacently on six short reinforced concrete columns and positioned beneath two, 110-kip (490-kN) capacity hydraulic actuators as shown in Figure 2a. Each column had a 2 in. (50.8 mm) steel reinforced elastomeric bearing pad to support the stems of the channel girders. The hydraulic actuators were placed 24 in. (610 mm) apart, each 12 in. (305 m) from mid-span, to create a nearly pure moment region. During controlled displacements of the actuators, the loads were measured using internal load cells in the actuator heads. As shown in Figure 2a, load was distributed by steel spreader beams to 3.0 in. (75 mm) diameter semi-circular load points centered over each stem of the east girder to create a four-point loading configuration.
Electronic clinometers (tiltmeters) were placed at mid-depth of the exterior face of each girder over each support to record girder rotation and measure the symmetry of the applied loads. String potentiometers were attached to the bottom of the interior stems at quarter points and at mid-span to measure vertical deflections of the girder. Two additional string potentiometers were placed at mid-span on the bottom of the east exterior stem.
Linear variable differential transformers (LVDTs) were used to measure axial displacements in the girders near mid-span, for use in calculating curvature of the girders. To record any gap openings in the shear key, LVDTs were also placed across both the top and bottom of the shear key joint at one quarter point and at mid-span.
Grouting Mixtures and Methods
UHPC and US Spec NA non-shrink grouts were used as grouting materials for these tests. The non-shrink grout mixture was proportioned according to the manufacturer’s recommendations.
During casting of each shear key, 4 in. (101.6 mm) cubes, a 4 × 8 in. (101.6 × 203.2 mm) cylinder and a 6 × 12 in. (152.4 × 304.8 mm) cylinder were cast for compressive strength testing and modulus of elasticity measurements. Before casting the shear keys, the bottom of the shear key joint was sealed with caulk to contain the grout until it had set.
To produce the improperly bonded UHPC shear key, the walls of the shear key were neither saturated with water nor intentionally roughened. After grouting, plastic was placed over the shear key to decrease evaporation and facilitate strength gain of the UHPC. After 24 h, the shear key was covered in wet burlap.
Before casting the US Spec NA non-shrink grout, the shear key walls were wetted to produce a more effective bond. After casting, the shear key was covered with plastic to prevent loss of moisture and aid strength gain. After 24 h, the plastic was removed and the shear key was covered with wet burlap.
For casting of the fully bonded UHPC shear key, the walls of the key were wetted to assist bond between the UHPC and precast concrete. Plastic was placed over the top of the shear key to prevent moisture loss. After 24 h, the plastic was replaced with wet burlap.
Cyclic Loading
Deflection limits from the AASHTO LRFD Bridge Design Specifications are summarized in Table 3 ( 16 ). These deflection limits include:
Criteria for deflections as outlined by AASHTO 2.5.2.6.2 which states an allowable deflection limit for vehicular loads on concrete bridges of Span/800 ( 16 ). For the clear span of 288 in. (7315 mm), the allowable deflection was calculated to be 0.36 in. (9.14 mm).
Application of a design truck as outlined by AASHTO 3.6.1.2.2 (
16
). The 72 in. (1830 mm) transverse spacing of wheels allows only a single wheel load from each axle to fit on the 914 mm width of a channel girder. Thus, a distribution factor of 1.0 was conservatively assumed for the following two configurations:
To produce the most extreme effect, the 32.0-kip (142-kN) axles were assumed to be spaced at 14.0 ft. (4.27 m), with one 16-kip (71.2-kN) wheel load from each of the two axles placed on the girder equidistant from mid-span. This loading configuration was used because the loads were applied by cylindrical steel load heads, not distributed over an area representative of a tire footprint. The mid-span deflection because of this configuration was calculated to be 0.228 in. (5.80 mm). The total force of a single 32.0-kip (142-kN) axle was assumed to act as a point load at mid-span. The mid-span deflection because of this configuration was calculated to be 0.388 in. (9.85 mm).
Summary of Deflection Criteria for Cyclic Loading
Cyclic Testing
The first full-scale test conducted was cyclic loading of the reinforced concrete channel girders with the shear key grouted with improperly bonded UHPC, subsequently referred to as Cyclic Loading 1 (CL1). The girder was first loaded at a constant rate of 0.05 in./min (1.27 mm/min) to a mid-span deflection of 0.40 in. (10.2 mm). Once the required actuator displacement was reached, the girder was unloaded, and the loading was repeated for a total of 1,000 load-unload cycles. A load rate of 2 cycles/min was used for the first 50 cycles to assess the test for any potential safety concerns. The load rate was then increased to 4 cycles/min for the remaining 950 cycles.
After completing CL1, the girders were separated and the UHPC grout was removed. The girders were then repositioned and prepared for the second test, which included wetting the shear key walls to ensure a more efficient bond with the grouting material. No roughening of the shear key surfaces was performed because test CL1 showed that it was not needed to maintain bond through the target deflections. In practice, roughening is needed to maintain bond through thermal variations that were not present in the laboratory during testing. The second test used US Spec NA non-shrink grout as the shear key grout. The same testing setup and loading sequence used for CL1 were used for this test. This test is referred to as Cyclic Loading 2 (CL2).
After completing CL2, the girders were once again separated and the grouting material was removed. At this point, a single girder was subjected to cyclic loading to quantify the behavior of the girder when no adjacent girder was available for load transfer. For this test, the east girder was instrumented as described in Figure 2. To remain consistent, cyclic loading of the single girder, referred to as Cyclic Loading 3 (CL3), was performed by loading the girder at a constant rate of 0.05 in./min (1.27 mm/min) to the desired deflection of 0.40 in. (10.2 mm) and then unloaded. Because only a single girder was tested in CL3, three load-unload cycles were adequate to observe the load-deflection behavior.
For the final full-scale test, the girders were once again placed adjacently and the shear key wetted to assist bonding of the UHPC grout. This test, referred to as Cyclic Loading 4 (CL4), used the same test setup described for CL1 and CL2 as shown in Figure 2. The girders were loaded to the desired deflection at a constant rate of 0.05 in./min (1.27 mm/min), unloaded, and then cycled for 100 load-unload cycles at 2 cycles/min. After completing the 100 cycles, deflection was increased to ultimate loading.
Ultimate Loading
Ultimate loading of the reinforced concrete channel girders was also as shown in Figure 2. The girder was loaded again to the service displacement of 0.40 in. (10.2 mm), from which loading was increased to a deflection of 0.8 in. (20.3 mm). The girder was then unloaded to the initial 0.40 in. (10.2 mm) of deflection. This process continued with an increase in maximum deflection of 0.40 in. (10.2 mm) after every unload cycle until a maximum deflection of 2.0 in. (50.8 mm) was reached. At this point, the girders were cycled between 0.40 in. (10.2 mm) and 2.0 in. (50.8 mm) of deflection for three cycles at a constant rate of 0.1 in./min (2.54 mm/min). After completing these cycles, the girders were unloaded and the residual deflection was measured. To complete the test, the girders were loaded to 0.40 in. (10.2 mm) of deflection, accounting for residual effects, and cycled for another 100 load-unload cycles at a rate of 4 cycles/min.
Results
Compressive Strength
Table 4 presents the average compressive strengths obtained from the UHPC and the substrate concrete. The minimum 28-day strength required for UHPC, according to ASTM C1856, is 120 MPa ( 7 ). The UHPC mixture used in this research met that requirement. The 28-day compressive strength for the substrate concrete was obtained from specimens that were moist-cured (73°F [23°C] and 98% relative humidity) to an age of 7 days and then cured at ambient conditions to an age of 28 days. This curing regimen was adopted to match the substrate curing used for the bond strength tests.
Compressive Strengths for Ultra-High-Performance Concrete (UHPC) and Substrate
Note: SD = standard deviation; psi = pounds per square inch.
Bond Strength
Slant Shear
Table 5 presents the average bond strengths from the slant-shear tests conducted at 7 and 28 days after UHPC was applied to the substrate. The American Concrete Institute (ACI) recommends a bond shear strength for slant shear of 1000 psi (6.90 MPa) for 7 days and 2000 psi (13.8 MPa) for 28 days ( 11 ). The results showed no clear trend of bond strength varying with surface roughness. This is because of the high variability of bond strength and small population of samples that were tested (three to six for each surface roughness at each age). However, each surface texture provided adequate bond strength at 7 days. It was also observed that shear strength increased from 7 to 28 days, except for the formed surface where it decreased by 23%. Although the 23% bond shear strength reduction from 7 days to 28 days for the formed specimens is large, a Student’s t-test produced a value of 0.51 indicating there is no statistical significance to this observation. Several of the shear strengths at 28 days did not meet the recommended shear strength of 2,000 psi (13.8 MPa). This is attributed to the low strength of the substrate concrete. Another possibility is that the substrate may have been compromised by cracks caused by chipping during surface preparation, although no evidence of such damage was observed. All of the slant-shear fractures with texture occurred in the substrate. However, the substrate was air-entrained, only wet-cured for 7 days, and was tested at 28 days. It appears that the strength at 28 days was not adequate for the supporting 2,000 psi (13.8 MPa) shear strengths.
Slant-Shear Bond Strengths psi (MPa)
Note: psi = pounds per square inch; SD = standard deviation.
Direct Tension
Table 6 presents the results from the direct tension tests conducted 7 and 28 days after the UHPC was applied to the substrate. ACI recommends a tensile strength for repair concrete at 7 and 28 days of 150 psi (1 MPa) and 250 psi (1.72 MPa), respectively ( 11 ). As shown in Table 6, the strength recommendations were met for the 7-day tests but not for the 28-day tests.
Direct Tensile Strengths
Note: psi = pounds per square inch; SD = standard deviation.
At 28 days, the formed surfaces experienced bond failures that indicate that the bond strength was inadequate. Additionally, the 28-day strengths of the formed surfaces were less than the strengths at 7 days. This observation is consistent with the similar decrease seen for formed surfaces in the slant-shear results. A Student’s t-test again showed that there was no statistical significance to this trend (t-test value of 0.45).
For the textured surfaces, all fractures occurred in the substrate. This is an indication that the bond was excellent, but tensile strength was limited by the strength of the substrate. As with the textured slant-shear specimens, substrate strength could have been reduced by cracks caused by chipping during surface preparation.
Shrinkage Effects
Shrinkage testing was conducted on 3 × 4 × 16 in. (76 × 102 × 406 mm) UHPC prisms. Tests were conducted for 28 days by monitoring length changes with a comparator. Figure 3 presents the results of the shrinkage tests. The specimens were cured for 6 days in the moist room, then removed and cured at ambient conditions for the remainder of the test. This curing method was implemented in Test 1 and repeated in Test 2. For comparison with the wet-cured specimens (W), ambient-cured specimens (D) were produced and tested in Test 2.
Average longer-term shrinkage results for ambient-cured specimens (D) and wet-cured specimens (W).
Shrinkage plateaued around 4 days for both Test 1 and Test 2. Once the specimens were removed from the moist room on day 7, shrinkage rate began to increase. Shrinkage plateaued again around 20 days, at approximately 450 µstrain for Test 1 and 420 µstrain for Test 2. Two plateaus also occurred in the results for ambient-cured specimens. The first plateau began on day 5 at around 320 µstrain and the second plateau occurred around day 20 at approximately 580 µstrain. The ambient-cured specimens experienced shrinkage of about 150–170 µstrain greater than the wet-cured specimens, which was expected.
Channel Girder Tests
Load Deflection Behavior
The full-scale tests were used to assess load-deflection behavior and shear and moment distribution of the assemblages. Deflections caused by four-point loading applied to a single girder (Figure 2) were used to assess load transfer across the shear key. To assess moment distribution, mean moments computed from CL3 were compared with moments from CL1, CL2, and CL4.
Cyclic Loading 1
CL1 utilized UHPC as the shear key grouting material without preparing the keyway for proper bonding of the grout. Therefore, the UHPC was considered improperly bonded. The east girder was loaded as at a mid-span deflection rate of 0.05 in./min (1.27 mm/min) until a deflection target of 0.40 in. (10.2 mm) was reached, and then cycled through 1,000 load-unload cycles. Figure 4, a and b , show the load-deflection behavior observed at mid-span. For the interior stems, the average difference in deflection between the east and west girders was found to be 0.017 in. (0.43 mm) (Figure 4a). However, the difference in deflection was greater between the exterior stem of the east girder and the interior stem of the west girder, where the difference was 0.084 in. (2.13 mm). This result reflects the lack of support on the east side of the loaded girder compared with the west side where the girders were joined. The small difference in deflection between the interior stems suggests the girders were deflecting nearly equally and transferring load adequately. The average peak load and mid-span deflection for CL1 were 13.9 kips (61.9 kN) and 0.41 in. (10.4 mm), respectively. The girders also appeared to remain linear elastic throughout the duration of CL1.
Full-scale testing results for load cases CL1, CL2, and CL4: (a) load versus deflection of east and west adjacent girders for CL1 (interior stems of each girder at midspan), (b) load versus deflection for CL1 (exterior stem of east girder and interior stem of west girder), (c) load versus deflection for CL2 (interior stems of each girder at mid-span), (d) load versus deflection for CL2 (exterior stem of east girder and interior stem of west girder), (e) load versus deflection for CL4 (interior stems of each girder at mid-span), and (f) load versus deflection for CL4 (exterior stem of east girder and interior stem of west girder).
Cyclic Loading 2
CL2 used US Spec NA non-shrink grout and was performed using the same methodology as CL1 to load the east girder to a mid-span deflection of 0.40 in. (10.2 mm) and then subject it to 1,000 load-unload cycles. Load deflection results for CL2 are presented in Figure 4, c and d . The average difference in deflection of the interior stems was found to be 0.015 in. (0.381 mm), shown in Figure 4c. In comparison, Figure 4d shows that the exterior stem of the east girder and the interior stem of the west girder had a difference in deflection of 0.081 in. (2.06 mm). When comparing these differential deflections with those measured during CL1, it was found that the differential displacement for CL1 was approximately 0.003 in. (0.08 mm) greater than for CL2. While the decreased deflections for CL2 may indicate slightly better load transfer across the shear key, this improvement is less than the resolution of the instrumentation (±0.015 in. [±0.381 mm]) and should not be considered significant.
The east girder for CL2 also had a greater exterior stem deflection compared with the interior stem, because of a lack of support and load transfer. The average peak load applied to the girders in CL2 was 13.4 kips (59.6 kN), with an average peak mid-span deflection of 0.40 in. (10.2 mm).
Cyclic Loading 3
The next cyclic test, CL3, was performed on a single girder with no grouted shear keys. The east girder alone was tested using the instrumentation and four-point longitudinal load configuration shown in Figure 2. The girder was loaded to a mid-span deflection of 0.39 in. (9.91 mm) and then subjected to three load-unload cycles. This test provided measurements of the load-deflection behavior of a single girder.
Figure 5 presents the load-deflection results recorded during CL3. Results from CL3 showed that an average load of 7.40 kips (32.9 kN) was required to induce a deflection of 0.40 in. (10.2 mm). This average is approximately half the recorded load from the CL1 and CL2 tests. This information was used to calculate the moment distribution factor. During the CL3 test, the girder behavior remained linear and no damage was observed.
Load versus deflection for CL3.
Cyclic Loading 4
The final full-scale test used UHPC as the grouting material and the shear key surfaces were wetted to ensure an effective bond. The east girder was again loaded to a mid-span deflection of 0.40 in. (10.2 mm) and then subjected to 100 load-unload cycles.
From Figure 4, e and f , the average deflection and load during CL4 were found to be 0.40 in. (10.2 mm) and 13.2 kips (58.7 kN), respectively. The differential deflection between interior stems was found to be 0.04 in. (1.02 mm) while the east exterior stem and west interior stem were found to have a difference in deflection of 0.14 in. (3.56 mm). This differential deflection (0.04 in. [1.02 mm]) is an increase compared with the differential deflections observed during tests CL1 (0.017 in. [0.43 mm]) and CL2 (0.015 in. [0.381 mm]) which seems to indicate that the properly bonded UHPC grout did not perform as well as the improperly bonded UHPC grout and the non-shrink grout. However, the magnitudes of the differences between the three cases was not substantially greater than the accuracy of the instrumentation (±0.015 in. [±0.381 mm]) so these observations do not necessarily provide strong evidence that the properly bonded UHPC was outperformed by the improperly bonded UHPC and non-shrink grout tests. As with tests CL1 and CL2, no cracking or de-bonding was observed during test CL4.
Ultimate Loading
The ultimate loading test was conducted using the CL4 channel girder assemblage. Loading began at an initial deflection of 0.40 in. (10.2 mm) and increased to a deflection of 0.80 in. (20.3 mm) then unloaded to 0.40 in. (10.2 mm). This process continued with the peak deflection increasing by 0.40 in. (10.2 mm) after every unload cycle until a deflection of 2.0 in. (50.8 mm) was reached. At this point, the girders were cycled between 0.40 in. (10.2 mm) and 2.0 in. (50.8 mm) of deflection for three cycles at a rate of 0.1 in./min (2.54 mm/min) then the residual deflection was measured on unload of the girder. To complete the test, the girders were loaded to 0.40 in. (10.2 mm) of deflection beyond the residual effects, and cycled for another 100 load-unload cycles at 4 cycles/min.
The overload behavior measurements are presented in Figure 6, a and b . During this portion of ultimate testing, the average peak deflection was 1.99 in. (50.5 mm). Figure 6, a and b , show that the girders began to yield at approximately 1.2 in. (30.5 mm) of deflection. Yielding was also observed visually as multiple flexural cracks began to appear and propagate in the nearly pure moment region at mid-span.
Full-scale testing results: (a) load versus deflection for ultimate loading (interior stems of each girder at mid-span), (b) load versus deflection for ultimate loading (exterior stem of east girder and interior stem of west girder), (c) load versus deflection for final 100 cycles after ultimate loading (interior stems of each girder at mid-span), and (d) load versus deflection for final 100 cycles after ultimate loading (exterior stem of east girder and interior stem of west girder).
The girders were then unloaded at 0.1 in./min (2.54 mm/min) and the residual deflection caused by the ultimate loading was recorded as 0.3 in. (7.62 mm). The girders were once again loaded at 0.1 in./min (2.54 mm/min) to a deflection of 0.40 in. (10.2 mm), past the residual deflection, giving a total displacement measurement of 0.7 in. (17.8 mm). The girders were then subjected to 100 more load-unload cycles. These final cycles were completed at the service deflection limit (0.40 in. [10.2 mm]) and were performed to mimic the load-deflection behavior of a bridge structure that remains in service after an overload situation.
Figure 6, c and d , present the load-deflection response for the final 100 load-unload cycles. The final 100 cycles had an average peak load of 13.7 kips (60.9 kN), causing an average deflection of 0.68 in. (12.3 mm), including the residual deflection. On completing the cycles, it was observed that flexural cracks had propagated at regular intervals across the mid-span of both girders. While the vertical displacement had increased, the girders still required approximately the same load to reach a deflection of 0.4 in. (10.2 mm) beyond the 0.3 in. (7.62 mm) of permanent deflection. Also, greater differential displacement between the exterior east girder stem and interior west girder stem was observed after applying the 57.2 kip (254 kN) load, indicating bond damage to the shear key that reduced shear and moment transfer and increased the deflection of the east girder that was carrying greater load.
Comparison of Loadings
The load-deflection behaviors for cyclic and ultimate loadings are summarized in Table 7. The grouting methods were analyzed by comparing load-deflection behavior across the four cyclic tests. This comparison showed that each grouting material behaved similarly, and each two-girder system nearly doubled the load required to induce the target deflection of a single girder. This result implies that there was little mechanical difference between the three shear key designs.
Summary of Load-Deflection Behavior for all Loadings
Conclusion
Based on the research conducted during this project, the following conclusions were drawn from this work:
Adequate bond strength was achieved at 7 days for slant-shear and direct tension tests. Low bond strengths occurred at 28 days with fractures that occurred in the substrate. The low strengths appear to be caused by the low strength of the substrate.
Approximately 450 µstrain of shrinkage was observed in the UHPC that is expected to contribute to shear stresses at the bonded interface.
Full-scale testing of channel girder assemblages with UHPC grouted shear keys showed that the bond between the UHPC and mature girders was able to withstand maximum permissible deflections for an equivalent bridge, even with formed shear key walls that provided improper bond for the UHPC grout.
The excellent bond between the UHPC grout and the girders occurred with no lateral restraint (post-tensioning). This indicates that bond strength between UHPC and mature substrate with a formed surface is not as great a concern as the bond strength results seemed to indicate.
Non-shrink grout slightly outperformed UHPC grout, indicating that UHPC grout does not provide a mechanical benefit. The differential deflection measurements that indicated that the non-shrink grout outperformed the UHPC grout were not substantially greater than the accuracy of the instrumentation. Consequently, this result is not considered strong evidence.
While there does not appear to be a mechanical benefit to using UHPC grout, it may still be a desirable choice for shear key rehabilitation applications based on durability and economic considerations.
Footnotes
Acknowledgements
The authors would like to thank the following organizations for their contributions to this project: Tran-SET for funding the project; New Mexico Department of Transportation for their support of UHPC research and collaboration; and BASF Chemical Company for their donations of silica fume and admixtures.
Author Contributions
The authors confirm contribution to the paper as follows: study conception and design: C. Newtson, B. Weldon; data collection: E. Flores, J. Varbel, W. Toledo; analysis and interpretation of results: E. Flores, J. Varbel, W. Toledo, C. Newtson, B. Weldon; draft manuscript preparation: E. Flores, J. Varbel, W. Toledo, C. Newtson, B. Weldon. All authors reviewed the results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Transportation Consortium of South-Central States (Tran-SET) as project 18CNMS01 “UHPC Shear Keys in Concrete Bridge Superstructures.”